Guile interpeter crash

Guile interpeter crash

[Dmitry Bogatov]

[-- Attachment #1: Type: text/plain, Size: 1136 bytes --]


Here is code that results crash (return 134). Hope it is interesting.

guile (GNU Guile) 2.0.9
Copyright (C) 2013 Free Software Foundation, Inc.

(define *define-count* 0)
(define-syntax define#
    (syntax-rules ()
	((_ obj ...)
	 (begin
	     (set! *define-count* (1+ *define-count*))
	     (define obj ...)))))
(define-syntax define
    (syntax-rules ()
	((_ obj ...)
	 (define# obj ...))))
(define# foo 2)
(display *define-count*)

$ guile --no-auto-compile crash.scm
Backtrace:
In ice-9/psyntax.scm:
  976:  19 Aborted
$ guile --auto-compile crash.scm
;;; note: auto-compilation is enabled, set GUILE_AUTO_COMPILE=0
;;;       or pass the --no-auto-compile argument to disable.
;;; compiling /share/home/kaction/build/crash.scm
;;; WARNING: compilation of /share/home/kaction/build/crash.scm failed:
;;; ERROR: Throw to key `vm-error' with args `(vm-run "VM: Stack overflow" ())'.
aborted

--
Best regards, Dmitry Bogatov <KAction@gnu.org>,
Free Software supporter and netiquette guardian.
	git clone git://kaction.name/rc-files.git --depth 1
	GPG: 54B7F00D
Html mail and proprietary format attachments are forwarded to /dev/null.

[-- Attachment #2: Type: application/pgp-signature, Size: 835 bytes --]

Fwd: Guile interpeter crash

[Panicz Maciej Godek]

[-- Attachment #1: Type: text/plain, Size: 1182 bytes --]

---------- Forwarded message ----------
From: Panicz Maciej Godek <godek.maciek@gmail.com>
Date: 2013/10/1
Subject: Re: Guile interpeter crash
To: Dmitry Bogatov <KAction@gnu.org>


2013/10/1 Dmitry Bogatov <KAction@gnu.org>

>
> Here is code that results crash (return 134). Hope it is interesting.
>

The code you gave is not a proper Scheme program.
C.f.
http://www.schemers.org/Documents/Standards/R5RS/HTML/r5rs-Z-H-8.html#%_sec_5.3,
excerpt:

Although macros may expand into definitions and syntax definitions in any
context that permits them, it is an error for a definition or syntax
definition to shadow a syntactic keyword whose meaning is needed to
determine whether some form in the group of forms that contains the
shadowing definition is in fact a definition, or, for internal definitions,
is needed to determine the boundary between the group and the expressions
that follow the group

In other words, you cannot redefine define. I suppose the interpreter falls
into infinite loop when it tries to substitute define# with define, and
then define -- with define#.
That being said, I don't think it's fortunate to use the # character within
a symbol.

Best regards,
M.

[-- Attachment #2: Type: text/html, Size: 2165 bytes --]

Re: Fwd: Guile interpeter crash

[Mark H Weaver]

Panicz Maciej Godek <godek.maciek@gmail.com> writes:

> ---------- Forwarded message ----------
> From: Panicz Maciej Godek <godek.maciek@gmail.com>
> Date: 2013/10/1
> Subject: Re: Guile interpeter crash
> To: Dmitry Bogatov <KAction@gnu.org>
>
> 2013/10/1 Dmitry Bogatov <KAction@gnu.org>
>
>     Here is code that results crash (return 134). Hope it is
>     interesting.
>
> The code you gave is not a proper Scheme program.
> C.f.
> http://www.schemers.org/Documents/Standards/R5RS/HTML/r5rs-Z-H-8.html#%_
> sec_5.3, excerpt:
>
> Although macros may expand into definitions and syntax definitions in
> any context that permits them, it is an error for a definition or
> syntax definition to shadow a syntactic keyword whose meaning is
> needed to determine whether some form in the group of forms that
> contains the shadowing definition is in fact a definition, or, for
> internal definitions, is needed to determine the boundary between the
> group and the expressions that follow the group
>
> In other words, you cannot redefine define.

It can be done safely using the module system, by calling it something
other than 'define' in the module, and then renaming it on export.  For
example, see:

  http://git.savannah.gnu.org/gitweb/?p=guile.git;a=blob;f=module/ice-9/curried-definitions.scm;h=fa369906ccd222b6aa2f3692f17638ca02eb2a41;hb=HEAD

In practice, there are ways to redefine 'define' within the same module
in Guile 2.0, but it depends on implementation details and might not be
future proof.

> That being said, I don't think it's fortunate to use the # character
> within a symbol.

Yes, it's probably best to avoid that character in symbols.  The fact
that it happens to work is a quirk of our reader, and is undocumented.

For the documented characters that are allowed in symbols, see:

http://www.gnu.org/software/guile/manual/html_node/Symbol-Read-Syntax.html

    Regards,
      Mark

Re: Fwd: Guile interpeter crash

[Dmitry Bogatov]

[-- Attachment #1: Type: text/plain, Size: 659 bytes --]


Mark H Weaver <mhw@netris.org> writes:
> It can be done safely using the module system, by calling it something
> other than 'define' in the module, and then renaming it on export.  For
> example, see:
>
>   http://git.savannah.gnu.org/gitweb/?p=guile.git;a=blob;f=module/ice-9/curried-definitions.scm;h=fa369906ccd222b6aa2f3692f17638ca02eb2a41;hb=HEAD
Thanks! I already found working solution, but your is cleaner.

--
Best regards, Dmitry Bogatov <KAction@gnu.org>,
Free Software supporter and netiquette guardian.
	git clone git://kaction.name/rc-files.git --depth 1
	GPG: 54B7F00D
Html mail and proprietary format attachments are forwarded to /dev/null.

[-- Attachment #2: Type: application/pgp-signature, Size: 835 bytes --]